CD4 percentage has primarily been used to determine when to initiate antiretroviral therapy (ART) in HIV-infected children because of the normal age-related decline in absolute CD4 cell count. However, both CD4 cell markers are used in current treatment guidelines, with ART initiation thresholds dependent on age [1–3]. Thus, there may be a dilemma about whether to initiate or defer ART when one marker is below the threshold but the other is above.
We evaluated the frequency of such discordance between CD4 cell count and CD4 percentage using data from a large group of HIV-infected children followed longitudinally in the era before effective (combination) ART was widely available. We also fitted models to assess whether CD4 cell count or percentage is the stronger predictor of the short-term risk of clinical disease progression. This may assist clinical decision-making when the two markers give conflicting indications about whether to start ART, as well as informing which CD4 marker should be targeted in efforts to develop simple, point-of-care technology to assess immune status in children in resource-limited settings.
HIV Paediatric Prognostic Markers Collaborative Study (HPPMCS) is a meta-analysis of individual data from children with perinatal HIV-1 infection from 17 European and US cohort studies and randomized trials, covering the period 1983–2002 [4,5]. We used the same inclusion and censoring criteria as in previous studies, including censoring 6 months after the initiation of any antiretroviral drug other than zidovudine monotherapy [4,5]. In addition, this analysis was limited to visits when both CD4 cell count and percentage were measured.
Several arbitrary definitions of discordance between CD4 cell count and percentage have been proposed in the adult literature [6–8]. CD4 discordance in this study was defined in relation to ART initiation thresholds in World Health Organization (WHO) and European Paediatric European Network for Treatment of AIDS (PENTA) treatment guidelines [1,2], when one marker indicates treatment should be started, whereas the other indicates treatment may be deferred (Table 1). Discordance could not be examined with reference to US guidelines as these recommend the use of CD4 percentage under age 5 years (treat if <25%) and CD4 cell count above age 5 years (treat if <350 cells/μl) .
Progression to AIDS  or death was investigated using Cox proportional hazards models (on an age-time scale) with time-dependent CD4 cell measurements; these were assumed to ‘expire’ after an interval of 12 months to account for occasional long gaps between measurements. Separate models were fitted for four age groups (<1 year, 1–2 years, 3–4 years, ≥5 years). The main objective of these models was to assess whether CD4 cell count contained prognostic information over and above that provided by CD4 percentage, and vice versa. Thus, for the former question, we allowed a flexible function for CD4 percentage (a cubic spline, with knots at the 10th, 50th, and 90th centiles ) and added a simple linear term to test the effect of CD4 cell count; for the latter question, the roles of CD4 cell count and percentage were reversed. To facilitate comparison of the effect of the two CD4 cell markers, which are measured on different scales, hazard ratios were plotted against age-group specific centiles (relative to the 50th centile). CD4 cell counts and percentages were square-root transformed to improve model fit. Analyses were performed in Stata (version 10.0; Stata Corp., College Station, Texas, USA).
A total of 3345 children, with 21 815 pairs of CD4 cell count and percentage measurements [median 5, interquartile range (IQR) 2–9, per child], were included in the analysis. Four thousand and twenty-three measurements were recorded in infants aged less than 1 year, 5496 at ages 1–2 years, 5007 at ages 3–4 years, 6173 at ages 5–9 years, and 1116 in children aged 10–16 years old. The median follow-up per child was 1.7 years (IQR 0.8–3.1). Approximately half (1717, 51%) of the children were female and 1205 (36%) were of black ethnicity.
Discordance between CD4 cell count and percentage
Overall, 14% of all measurements were classified as discordant by WHO guidelines and 21% by European guidelines. However, as discordance is primarily of practical relevance when children become potentially eligible for treatment, the main analysis is based on the first measurement per child at which one or both markers fell below the treatment thresholds (Table 1). Discordance was common at all ages for both WHO and European guidelines. Under European guidelines, 55% of measurements were discordant, with CD4 percentage more frequently below the treatment threshold (33%) than CD4 cell count (22%). Under WHO guidelines a similar pattern was observed for these age groups (1–4 years), whereas in the below 1 year age group a much higher proportion of measurements were discordant due to a low CD4 cell count as opposed to a low CD4 percentage (45 vs. 17%).
We also examined the discordance pattern at the visit following the one when of the markers first fell below the treatment thresholds [median (IQR) interval between visits: 2.8 (1.8, 4.1) months]. Under WHO guidelines, when only CD4 cell count fell below the threshold, both markers were subsequently above the threshold for 275 of 517 (53%) visits, whereas for 130 of 517 (25%) the same pattern of discordance was observed. The corresponding values for CD4 percentage were 235 of 485 (49%) and 155 of 485 (32%). Similar results were obtained under European guidelines (not shown).
Clinical disease progression
Nine hundred and eighty (29%) children either developed AIDS or died. Figure 1 shows the estimated hazard ratios for progression to AIDS/death for the effect of CD4 percentage controlling for CD4 cell count and the effect of CD4 cell count controlling for CD4 percentage. Findings were consistent across the four age groups analysed: CD4 cell count added significant predictive information given knowledge of CD4 percentage; and virtually all information about the risk of short-term disease progression was captured by CD4 cell count, with minimal additional contribution from CD4 percentage. For example, the estimated hazard ratio for AIDS/death at the 10th centile of CD4 cell count (relative to the 50th centile) was 2.3 [95% confidence interval (CI) 1.8, 2.8] at age below 1 year, 2.2 (1.4, 3.0) at age 1–2 years, 2.4 (1.3, 3.5) at 3–4 years, and 2.7 (0.7, 4.6) at age at least 5 years. In contrast, the corresponding values for CD4 percentage were only 0.9 (0.7, 1.0), 1.2 (0.8, 1.6), 1.2 (0.6, 1.9), and 1.4 (0.4, 2.5).
Finally, the models were re-fitted to a restricted dataset in which data were censored after the first visit when both CD4 cell count and percentage fell below the respective WHO treatment thresholds. The rationale for this analysis was to exclude follow-up when all children would now presumably receive treatment. Findings, based on 656 events, were similar to those from the main analysis (not shown).
The high risk of rapid disease progression during infancy and the results of the CHER trial [5,11] have led to a general consensus that all HIV-infected infants should receive ART [2,3]. In addition, recent revisions of US and European guidelines advocate that the immunological monitoring of children aged at least 5 years should be based on CD4 cell count, partly because older children experience similar disease progression rates to young HIV-infected adults for the same value of CD4 cell count . However, treatment guidelines are not always strictly followed and, for completeness, we have presented comparisons of CD4 cell count and percentage at all ages.
The two CD4 cell markers were frequently discrepant, occurring in over one-half of children as they crossed one of the treatment initiation thresholds. However, both markers are subject to high within-patient variability and measurement error , and we found that the marker which was originally below the threshold had often ‘normalized’ at the next visit. This highlights the importance of repeating immunological tests before making clinical decisions which are not easily reversible. Nevertheless, repeatedly discordant CD4 cell count and CD4 percentage values were observed in some children, raising the question of which marker to place more emphasis on. A conservative approach favoured by some paediatricians is to initiate treatment if either CD4 cell marker is below the threshold. The presence of HIV-related clinical symptoms or an elevated HIV RNA viral load is also among the factors to be considered in the decision whether to start ART [1–3].
We found, however, that CD4 percentage has no or little prognostic value over and above that contained in CD4 cell count, regardless of age. This finding has been confirmed in some, but not all, studies conducted in HIV-infected adults [6–8], and is somewhat counter-intuitive in view of the greater stability of CD4 percentage (which is less affected by the presence of other infections, for example). The practical implication is that a CD4 cell count below a treatment threshold (when CD4 percentage is above the threshold) provides a stronger impetus for initiating ART than vice versa. A difficulty with using CD4 cell count, however, is the substantial variation in disease progression risk for a given level of this marker, even with relatively narrow age bands. For example, the estimated 12-month risk of AIDS or death at the European guideline treatment initiation threshold varies between 23 and 5% in the 1–2 years age band (1000 cells/μl) and between 8 and 3% in the 3–4 years age band (500 cells/μl) . This suggests a possible advantage in considering age as a continuous variable as is the case for paediatric anthropometric charts .
Because our study was set up to examine paediatric prognostic markers in the absence of effective ART, it is important to note that the data were collected in the late 1980s and early 1990s. Technology to enumerate lymphocyte subsets has subsequently evolved, particularly the replacement of dual-platform procedures by single-platform procedures, allowing determination of absolute CD4 cell count directly by flow cytometry , although a strong correlation between dual and single-platform procedures has been reported in other studies . We also reiterate that the level and direction of discordance we have reported depends closely, by definition, on the specific values of the treatment initiation thresholds.
Our findings also inform the debate on CD4 cell monitoring in resource-limited settings, since the cheaper and simpler technologies which have been developed usually estimate CD4 cell count but not CD4 percentage . The fact that the former marker is a more powerful prognostic marker than the latter is encouraging for the use of these alternative technologies. However, mortality levels, controlling for age and CD4 cell count or percentage, are much higher among untreated HIV-infected children in Africa compared with Europe and North America  and the generalizability of these results to an African population is unknown.
Steering and Writing Committee: K. Boyd, D.T. Dunn, H. Castro, D.M. Gibb, T. Duong (Medical Research Council Clinical Trials Unit, London, UK); J.P. Aboulker (INSERM SC10, Villejuif); M. Bulterys (Division of HIV/AIDS Prevention, Centers for Disease Control and Prevention, Atlanta, USA; Perinatal AIDS Collaborative Transmission Study); M. Cortina-Borja (Institute of Child Health, University College London; European Collaborative Study); C. Gabiano (Department of Pediatrics, University of Turin, Turin, Italy; Italian Register for HIV Infection in Children); L. Galli (Department of Pediatrics, University of Florence, Florence, Italy; Italian Register for HIV Infection in Children); C. Giaquinto (Department of Pediatrics, University of Padova, Padova, Italy; Paediatric European Network for Treatment of AIDS [PENTA]); D.R. Harris [Westat, Rockville, MD, USA; National Institute of Child Health and Human Development (NICHD) Intravenous Immunoglobulin Study Group]; M. Hughes [Harvard School of Public Health, Boston, MA, USA; Pediatric AIDS Clinical Trials Group (PACTG)]; R. McKinney (Duke University Medical Center, Durham, NC, USA; PACTG); L. Mofenson (National Institute of Child Health and Human Development, National Institutes of Health, Rockville, MD, USA; NICHD Intravenous Immunoglobulin Study Group); J. Moye (National Institute of Child Health and Human Development; Women and Infants Transmission Study); M.L. Newell (Institute of Child Health, University College London; European Collaborative Study); S. Pahwa (North Shore-LIJ Research Institute, Manhasset, NY, USA; PACTG); P. Palumbo (UMDNJ Medical School, Newark, NJ, USA; Perinatal AIDS Collaborative Transmission Study); C. Rudin (University Children's Hospital, Basel, Switzerland; Swiss Mother and Child HIV Cohort Study); M. Sharland [St George's Hospital Medical School, London, UK; Collaborative HIV Paediatric Study (CHIPS) of UK and Ireland]; W. Shearer (Baylor College of Medicine, Houston, TX, USA; Pediatric Pulmonary and Cardiovascular Complications of HIV Infection Study); B. Thompson (Clinical Trials and Surveys Corp, Baltimore, MD, USA; Women and Infants Transmission Study); P. Tookey (Institute of Child Health, University College London; CHIPS).
1. Antiretroviral therapy of HIV infection in infants and children in resource-limited settings: towards universal access. Recommendations for a public health approach. World Health Organisation; 2006.
4. HIV Paediatric Prognostic Markers Collaborative Study Group. Predictive value of absolute CD4 cell count for disease progression in untreated HIV-1-infected children. AIDS 2006; 20:1289–1294.
5. HIV Paediatric Prognostic Markers Collaborative Study. Short-term risk of disease progression in HIV-1-infected children receiving no antiretroviral therapy or zidovudine monotherapy: a meta-analysis. Lancet 2003; 362:1605–1611.
6. Centers for Disease Control and Prevention. 1994 revised classification system for human immunodeficiency virus infection in children less than 13 years of age. MMWR 1994; 43 (No. RR-03):1–21.
7. Gebo KA, Gallant JE, Keruly JC, Moore RD. Absolute CD4 vs. CD4 percentage for predicting the risk of opportunistic illness in HIV infection. J Acquir Immun Defic Syndr 2004; 36:1028–1033.
8. Goicoechea M, Haubrich R. CD4 lymphocyte percentage versus absolute CD4 lymphocyte count in predicting HIV disease progression: an old debate revisited. J Infect Dis 2005; 192:945–947.
9. Hulgan T, Shepherd BE, Raffanti SP, Fusco JS, Beckerman R, Barkanic G, Sterling TR. Absolute count and percentage of CD4+ lymphocytes are independent predictors of disease progression in HIV-infected persons initiating highly active antiretroviral therapy. J Infect Dis 2007; 195:425–431.
10. Durrleman S, Simon R. Flexible regression models with cubic splines. Stat Med 1989; 8:551–561.
11. Violari A, Cotton MF, Gibb DM, Babiker AG, Steyn J, Madhi SA, et al, for the CHER study team. Early antiretroviral therapy and mortality among HIV-infected infants. N Engl J Med 2008; 359:2233–2244.
12. Dunn D, Woodburn P, Duong T, Peto J, Phillips A, Gibb D, Porter K, for the HIV Paediatric Prognostic Markers Collaborative Study and the Concerted Action on Sero-Conversion to AIDS and Death in Europe (CASCADE) Collaboration. Current CD4 cell count and the short-term risk of AIDS and death before the availability of effective antiretroviral therapy in HIV-infected children and adults. J Infect Dis 2008; 197:398–404.
13. Hughes MD, Stein DS, Gundacker HM, Valentine FT, Phair JP, Volberding PA. Within-subject variation in CD4 lymphocyte count in asymptomatic human immunodeficiency virus infection: implications for patient monitoring. J Infect Dis 1994; 169:28–36.
14. Freeman JV, Cole TJ, Chinn S, Jones PR, White EM, Preece MA. Cross sectional stature and weight reference curves for the UK, 1990. Arch Dis Child 1995; 73:17–24.
15. Paintsil E, Ghebremichael M, Romano S, Andiman WA. Absolute CD4+ T-lymphocyte count as a surrogate marker for pediatric human immunodeficiency virus disease progression. Pediatr Infect Dis J 2008; 27:629–635.
16. Jeganathan S, Bansal M, Smith DE, Gold J. Comparison of different methodologies for CD4 estimation in a clinical setting. HIV Med 2008; 9:192–195.
17. Diabouga S, Chazallon C, Kazatchkine MD, Van de Perre P, Inwolev A, M'Boup S, et al. Successful implementation of a low-cost method for enumerating CD4+ T Lymphocytes in resource limited settings: the ANRS 12–26 study. AIDS 2003; 17:2201–2208.
18. Cross Continents Collaboration for Kids (3Cs4kids) Analysis and Writing Committee. Markers for predicting mortality in untreated HIV-infected children in resource-limited settings: a meta-analysis. AIDS 2008; 22:97–105.
© 2010 Lippincott Williams & Wilkins, Inc.